combatseq_batchcorrect_withfilter

library(tidyverse)
library(ggplot2)
library(rtracklayer)
library(AnnotationDbi)
library(EnsDb.Hsapiens.v86)
library(edgeR)
library(ggrepel)

load("~/mccoyLab/collabs/doubleseq_2021/code/batch_correct/combatseq_corrected_withshrinkage.RData")

#Importing metadata to differentiate between pregnant and not pregnant and other main covariates
meta <- read.csv("~/mccoyLab/collabs/doubleseq_2021/tidied_meta/tidied_meta_CREATE_kw_20220308.csv", row.names = 1) %>% 
  as.data.frame() %>% 
  mutate(across(c("cDNA_RT_Date", "Library_Prep_Date", "Sequencing_Date", "Study_Participant_ID"), as.factor))
#subsetting meta data to just the samples with sequencing data 
remove <- setdiff(rownames(meta), colnames(corrected_jg_rts))
if (length(which(rownames(meta) %in%  remove)) > 0){
  meta <- meta[-which(rownames(meta) %in%  remove),]
}
remove

## [1] "66-D-3-L"   "93-A-1-L"   "109-B-7-L"  "127-A-1-L"  "144-A-3-L" 
## [6] "154-A-1-L"  "155-A-16-L" "169-A-2-L"  "197-A-3-L"

counts_order <- order(colnames(corrected_jg_rts))
meta_order <- order(rownames(meta))
counts <- corrected_jg_rts[, counts_order]
meta <- meta[meta_order, ]
meta$name <- rownames(meta)

counts_filter <- counts[rowMeans(counts) > 7.63295995515296,]

logged_cf <- log2(counts_filter + 1) %>% as.data.frame()
data_long_logged_cf <- logged_cf %>%
  pivot_longer(colnames(logged_cf)) %>% 
  as.data.frame()
with_batch_lcf <- left_join(data_long_logged_cf, meta[,c("name", "cDNA_RT_Date", "Sequencing_Date", "Pregnant")], by="name")
ggplot(data_long_logged_cf, aes(x=value, col=name)) + geom_density(alpha=0.2) + theme_bw() + theme(legend.position = "none", panel.background = element_blank(), panel.grid = element_blank())

ggplot(with_batch_lcf, aes(x=value, group_by = name, col=as.factor(Pregnant))) + geom_density(alpha=0.2) + theme_bw() + theme(panel.background = element_blank(), panel.grid = element_blank())

ggplot(with_batch_lcf, aes(x=value, group_by = name, col=as.factor(Pregnant))) + geom_density(alpha=0.2) + theme_bw() + theme(panel.background = element_blank(), panel.grid = element_blank()) + facet_wrap(~cDNA_RT_Date)

ggplot(with_batch_lcf, aes(x=value, group_by = name, col=as.factor(Pregnant))) + geom_density(alpha=0.2) + theme_bw() + theme(panel.background = element_blank(), panel.grid = element_blank()) + facet_wrap(~Sequencing_Date)

pca_res_cf <- prcomp(t(counts_filter))
var_explained_cf <- pca_res_cf$sdev^2/sum(pca_res_cf$sdev^2)

pca_res_cf$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Sequencing_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_cf[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_cf[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_cf$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=meta$Sequencing_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_cf[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_cf[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_cf$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$cDNA_RT_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_cf[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_cf[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_cf$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=meta$cDNA_RT_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_cf[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_cf[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_cf$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Pregnant))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_cf[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_cf[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_cf$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=as.factor(meta$Pregnant))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_cf[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_cf[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

file_gencode <- "~/genomes/hg38_genome/gencode.v34.annotation.gtf"
gtf <- rtracklayer::import(file_gencode) %>% 
  as.data.frame() %>% 
  dplyr::filter(type == "gene") %>% 
  dplyr::select(gene_id, seqnames, width) %>% 
  dplyr::rename(ensembl_gene_id = gene_id) %>% 
  dplyr::rename(chromosome_name = seqnames) %>% 
  dplyr::rename(length = width)
gene_table <- gtf[match(rownames(counts_filter), gtf$ensembl_gene_id),]

dgeFullData <- DGEList(counts_filter)
#normalize counts by TMM
TMMFullData <- calcNormFactors(dgeFullData, method="TMM")
TMMCounts <- as.matrix(TMMFullData$counts)

#TPM Calculation
calc_tpm <- function(x, gene.length) {
  x <- as.matrix(x)
  len.norm.lib.size <- colSums(x / gene.length)
  return((t(t(x) / len.norm.lib.size) * 1e06)/ gene.length)
}

#creates a matrix with calculated TPM values for each sample from TMM normalized counts and the gene lengths
rawTPMvals <- calc_tpm(TMMFullData, gene.length = gene_table$length)

pca_res_tmm <- prcomp(t(TMMCounts))
var_explained_tmm <- pca_res_tmm$sdev^2/sum(pca_res_tmm$sdev^2)

pca_res_tpm <- prcomp(t(rawTPMvals))
var_explained_tpm <- pca_res_tpm$sdev^2/sum(pca_res_tpm$sdev^2)

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Sequencing_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=meta$Sequencing_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_tmm[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$cDNA_RT_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=meta$cDNA_RT_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_tmm[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Pregnant))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=as.factor(meta$Pregnant))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_tmm[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Sequencing_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=meta$Sequencing_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_tpm[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$cDNA_RT_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=meta$cDNA_RT_Date)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_tpm[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Pregnant))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=as.factor(meta$Pregnant))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_tpm[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

tmmsum <- colSums(TMMCounts)
rawsum <- colSums(counts_filter)

pca_res_cf$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=rawsum)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_cf[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_cf[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_cf$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=rawsum)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_cf[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_cf[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=tmmsum)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC2, x=PC3, color=tmmsum)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%"),
       x=paste0("PC3: ", round(var_explained_tmm[3]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.integer(meta$Study_Participant_ID))) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$lining_thickness_mm)) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Oocyte_Age)) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Uterus_Age)) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Sperm_Age)) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Embryo_grade_at_freezing))) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, label=rownames(meta))) + 
  geom_point() + 
  geom_text_repel() +
  theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(panel.background = element_blank(), panel.grid = element_blank())

oi1 <- "163-A-1-L" 
oi1i <- which(rownames(meta) == oi1)
oi2 <- "205-A-5-L"
oi2i <- which(rownames(meta) == oi2)
coloi <- rep(0, length(rownames(meta)))
coloi[c(oi1i, oi2i)] <- 1
pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color = as.factor(coloi), label=rownames(meta))) + 
  geom_point() + 
  geom_text_repel() +
  theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(panel.background = element_blank(), panel.grid = element_blank())

sample_vec <- c("101-A-10-L", "103-A-4-L","104-A-8-L",  "107-C-3-L","108-A-3-L", "11-B-12-L", "122-A-4-L",  "125-A-4-L", "129-A-3-L", "130-A-13-L", "15-C-4-L", "152-A-5-L", "156-A-2-L",  "167-C-4-L", "173-B-5-L", 
"179-A-6-L", "180-A-1-L",  "185-A-2-L", "2-A-3-L",    "202-A-11-L","218-A-5-L",
"31-A-1-L","36-A-2-L",   "37-A-4-L","4-B-1-L",   
"54-A-10-L","59-B-6-L",   "73-A-2-L","75-A-4-L",   "88-A-7-L","89-A-2-L",   "99-A-3-L" )

oii <- which(rownames(meta) %in% sample_vec)
coloi <- rep(0, length(rownames(meta)))
coloi[oii] <- 1
pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color = as.factor(coloi), label=rownames(meta))) + 
  geom_point() + 
  geom_text_repel() +
  theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(panel.background = element_blank(), panel.grid = element_blank())

## Warning: ggrepel: 140 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$AOD)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$GC)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$BMI)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Infertility_type)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Previous_pregnancy)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Past_surgical_hist)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$FSH_D3)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$AMH)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$AFC)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$LH)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$E2_Day_2_3)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Type_of_trigger))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Total_Gn)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$E2_on_trigger)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Num_eggs)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Num_MII)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Time_to_stripping_hr)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Time_to_ICSI_hr)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Time_to_Biopsy_hr)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Interpretation)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Natural_cycle))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Medication))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Prednisone))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Fragmin))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$EG))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$HCG))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$IL))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$GCSF))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Metformin))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$PRP))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Sildenafil))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Transferring_physician))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Transfer_catheter_used))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_SSM_GC))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_Egg_factor))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_MF))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_Uterine_factor))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_TF))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_RPL))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_RIF))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_Unexplained))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$PMdH_none))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$PMdH_vasculitis))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$PMdH_immune))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$PMdH_stress_hormones))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Ongoing_pregnancy))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tmm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Final_outcome))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tmm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tmm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.integer(meta$Study_Participant_ID))) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$lining_thickness_mm)) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Oocyte_Age)) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Uterus_Age)) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Sperm_Age)) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Embryo_grade_at_freezing)) + 
  geom_point() + theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, label=rownames(meta))) + 
  geom_point() + 
  geom_text_repel() +
  theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(panel.background = element_blank(), panel.grid = element_blank())

oi1 <- "163-A-1-L" 
oi1i <- which(rownames(meta) == oi1)
oi2 <- "205-A-5-L"
oi2i <- which(rownames(meta) == oi2)
coloi <- rep(0, length(rownames(meta)))
coloi[c(oi1i, oi2i)] <- 1
pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color = as.factor(coloi), label=rownames(meta))) + 
  geom_point() + 
  geom_text_repel() +
  theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(panel.background = element_blank(), panel.grid = element_blank())

oii <- which(rownames(meta) %in% sample_vec)
coloi <- rep(0, length(rownames(meta)))
coloi[oii] <- 1
pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color = as.factor(coloi), label=rownames(meta))) + 
  geom_point() + 
  geom_text_repel() +
  theme_bw() +
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(panel.background = element_blank(), panel.grid = element_blank())

## Warning: ggrepel: 130 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$AOD)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$GC)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$BMI)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Infertility_type)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Previous_pregnancy)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Past_surgical_hist)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$FSH_D3)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$AMH)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$AFC)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$LH)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$E2_Day_2_3)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Type_of_trigger))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Total_Gn)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$E2_on_trigger)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Num_eggs)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Num_MII)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Time_to_stripping_hr)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Time_to_ICSI_hr)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Time_to_Biopsy_hr)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=meta$Interpretation)) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Natural_cycle))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Medication))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Prednisone))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Fragmin))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$EG))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$HCG))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$IL))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$GCSF))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Metformin))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$PRP))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Sildenafil))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Transferring_physician))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Transfer_catheter_used))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_SSM_GC))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_Egg_factor))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_MF))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_Uterine_factor))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_TF))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_RPL))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_RIF))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$InfD_Unexplained))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$PMdH_none))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$PMdH_vasculitis))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$PMdH_immune))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$PMdH_stress_hormones))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Ongoing_pregnancy))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())

pca_res_tpm$x %>% 
  as.data.frame %>% 
  ggplot(aes(y=PC1, x=PC2, color=as.factor(meta$Final_outcome))) + 
  geom_point() + theme_bw() + 
  labs(y=paste0("PC1: ", round(var_explained_tpm[1]*100,1), "%"),
       x=paste0("PC2: ", round(var_explained_tpm[2]*100,1), "%")) +
  theme(legend.position = "top", panel.background = element_blank(), panel.grid = element_blank())